1. Field of the Invention
This invention relates generally to signal processing, and more particularly to conversion of analog signals to digital signals.
2. Description of the Related Art
Conversion of analog signals to digital signals is employed for many signal processing applications. Examples of such applications include communications, sonar, radar, signals intelligence high quality headsets, hearing devices, etc. In some applications, a strong interfering signal may be present along with a weak desired analog signal that is to be processed by an analog to digital converter (ADC or A2D). Under such conditions, the ADC may not have enough dynamic range to handle the strong interferer and weak signal simultaneously. In this case, the ADC may become saturated by the strong interfering signal and the weak desired signal may be at best distorted and hard to detect and measure, and at worst may be rendered completely non-recoverable.
In the past, more bits have been added to an ADC in an attempt to increase dynamic range and facilitate analog to digital conversion of weaker desired signals in environments where strong interfering signals are present. However, increasing the number of bits results in limited bandwidth or higher power consumption, large form factor, heat dissipation, and cost. Furthermore, the current state of the art limits the amount of dynamic range that can be achieved by adding-bits to an ADC. Another past approach for addressing strong interference and to increase the effective dynamic range has been to add upstream programmable analog filtering, together with interference detection and estimation, and controls for the programmable analog filtering. However, the interference suppression capability of this approach is limited by the shape control of analog filters, and it is not possible with this approach to remove an interferer that is occupying the same spectral space as the weak signal. It has also been proposed to address strong interference with weak analog signals by using a noise-shaping delta sigma ADC to suppress the interference while passing the desired weaker signal. While presenting an integrated approach with better form factor, this proposal suffers from similar drawbacks as upstream analog filtering. Further, filter performance and flexibility provided by currently available noise-shaping ADCs is not as good as the filter performance and flexibility provided by available separate analog filters.
Another proposed technique to effectively increase the dynamic range of an ADC is to use non-uniform sampling at an average rate less than Nyquist sampling. The advantage of non-uniform sampling is that the ADC average speed is slower and hence the ADC can be designed with more bits while keeping the total cost and power consumption reasonable and allowing more effective bits to be achieved. This approach suffers from lost information caused by the non-uniform sub-Nyquist sampling (limiting the signal environment that the technique will work) and costly post-processing requirements, along with difficulty in achieving the timing accuracy required for signal re-construction even in the limited environments that the technique can work.
Another technique used to increase the dynamic range of ADCs is to time-interleave multiple ADCs. This allows slower ADCs with more bits to be used, while increasing the total sample rate by using multiple ADCs. This technique suffers from increased power consumption, heat dissipation, and costly post-processing to correct interleaving imperfections and to compensate for differing channel characteristics. In addition, this technique is limited as to how many ADCs effectively be interleaved in practice, even with post-processing correction.
Yet another proposal has been made to combine the outputs of two ADCs that process an input signal that includes a weak desired signal in the presence of a strong interferer. In this latter proposal, a first one of the two ADCs is provided with an attenuator that provides an attenuated input to the first ADC, and the second one of the two ADCs has a non-attenuated input. The second non-attenuated ADC becomes saturated in the presence of the strong interferer while the first attenuated ADC remains unsaturated. However, once the second non-attenuated ADC becomes saturated, there is not a way to intelligently combine the output information from the saturated second ADC with the output information from the non-saturated first ADC; thus this approach requires two separate receiver paths and the associated processing and is therefore inefficient.